GitDealFlowsignals

NeurIPS 2020 · 2020

Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks

Patrick Lewis, Ethan Perez, Aleksandara Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal

Facebook AI Research

What this paper is

Retrieval-Augmented Generation (Lewis et al., NeurIPS 2020) combined a seq2seq model with a non-parametric memory: a dense passage retrieval index over Wikipedia that the generator consults before answering. RAG matched or beat fine-tuned baselines on knowledge-intensive tasks while letting each answer cite the documents that informed it.

Abstract summary

Introduces Retrieval-Augmented Generation (RAG): an architecture that combines a pretrained sequence-to-sequence model (BART) with a non-parametric memory (a Dense Passage Retrieval index over Wikipedia). Demonstrates strong performance on knowledge-intensive NLP tasks while providing transparency about which documents informed each generation. Establishes the design pattern of retrieving documents before generating.

Our summary in our own words, see the canonical source links below for the original abstract.

Why we cite this paper

RAG is the architecture behind essentially every enterprise LLM deployment in 2026. The retrieval layer, typically using embedding models plus vector databases (Pinecone, Weaviate, Qdrant, Milvus), is one of the densest engineering-acceleration surfaces we track at /sector/database and /trend/ai-native-databases-2026.

Where this matters for deal flow

Key findings

  • 1Combining parametric (model weights) and non-parametric (retrieved documents) memory outperforms either alone on knowledge-intensive tasks.
  • 2Retrieval enables citation transparency, the model's outputs can be traced to specific source documents.
  • 3RAG addresses three core LLM limitations: knowledge cutoff dates, hallucination on out-of-distribution facts, and inability to cite sources.
  • 4The design pattern became the foundation of the modern vector-DB-backed LLM application stack.

Canonical sources

Related glossary terms

Frequently Asked Questions

What is RAG?

Retrieval-Augmented Generation, an architecture where a model retrieves relevant documents from an external knowledge store before generating an answer. See /define/rag for the full term definition.

Who wrote the RAG paper?

Patrick Lewis and colleagues at Facebook AI Research (FAIR), published at NeurIPS 2020 (arXiv:2005.11401).

Why is RAG used in enterprise LLM applications?

It grounds generations in retrieved documents, which mitigates knowledge-cutoff staleness, reduces hallucination on out-of-distribution facts, and enables source citation, the three properties most enterprise deployments require.

What infrastructure does RAG depend on?

An embedding model plus a vector database (Pinecone, Weaviate, Qdrant, Milvus) for the retrieval layer. We track the engineering acceleration of that surface at /sector/database and /trend/ai-native-databases-2026.

See who is building on this, before the round prices it in

We track engineering acceleration across the AI & Machine Learning and AI Infrastructure sectors this paper informs: commit velocity, contributor influx, and repo-creation pulse, surfacing breakout teams 21 to 47 days before the fundraise is public.

Five breakout startups, every Sunday, before the round gets crowded

The free Acceleration Watch: five venture-backed teams accelerating on the engineering signal, translated into plain English, 21 to 47 days before the deck circulates. No code-reading, no card.

Signed The Data Nerd · pseudonymous narrator · methodology over personality

Other research papers

Read our own methodology paper

Code-Side Sourcing methodology, replicable on the open dataset.

Read /methodology

Related papers

🚀 Explore Our Network

21-47 days
Signal Lead Time (median 31d)
$80M+
Rounds Tracked
90 sec
Per Scan
5,000+
Founders Tracked

One missed signal is a missed round. Get the Velocity Verdict in your inbox every Sunday free.

Get Free Signals

Free weekly digest. Cancel anytime. No spam, no VC pitches just data.